Comparitive mortality levels among selected … · cantly exceeds that of males above age ﬁve. Wild-born animals do not have mortality that differs signiﬁcantly from captive-born

Demographic Research a free, expedited, online journalof peer-reviewed research and commentaryin the population sciences published by theMax Planck Institute for Demographic ResearchKonrad-Zuse Str. 1, D-18057 Rostock · GERMANYwww.demographic-research.org

DEMOGRAPHIC RESEARCH

VOLUME 15, ARTICLE 14, PAGES 413-434PUBLISHED 17 NOVEMBER 2006http://www.demographic-research.org/Volumes/Vol15/14/DOI: 10.4054/DemRes.2006.15.14

Research Article

Comparative mortality levels amongselected species of captive animals

Iliana V. Kohler

Samuel H. Preston

Laurie Bingaman Lackey

c© 2006 Kohler et al.

This open-access work is published under the terms of the Creative CommonsAttribution NonCommercial License 2.0 Germany, which permits use,reproduction & distribution in any medium for non-commercial purposes,provided the original author(s) and source are given credit.See http://creativecommons.org/licenses/by-nc/2.0/de/

Table of Contents

1 Introduction 414

2 Description of data and analytic scheme 414

3 Results 4193.1 Life tables for groups of species 4193.2 Mortality variation by species, sex, and birth type 4213.3 Life table parameters for individual species 426

4 Discussion 430

Demographic Research – Volume 15, Article 14

research article

Comparative mortality levels among selectedspecies of captive animals

Iliana V. Kohler 1

Samuel H. Preston 2

Laurie Bingaman Lackey 3

Abstract

We present life tables by single year of age and sex for groups of animals and for 42individual mostly mammalian species. Data are derived from the International SpeciesInformation System. The survivorship of most of these species has never been mappedsystematically. We demonstrate that, in most of the groups, female survivorship signifi-cantly exceeds that of males above age five. Wild-born animals do not have mortality thatdiffers significantly from captive-born animals. While most species have mortality thatrises with age above the juvenile stage, there are several groups for which the age patternof mortality is nearly level.

1Corresponding author; Postdoctoral Fellow and Research Associate, Population Studies Center, Universityof Pennsylvania, 3718 Locust Walk, Philadelphia, PA 19104-6299, USA; Tel.: +1-(215)-898-7990; Fax:+1-(215)-898-2124; Email: [email protected].

2Fredrick J. Warren Professor of Demography, 3718 Locust Walk, University of Pennsylvania, Philadelphia,PA 19104-6299, USA; Email: [email protected].

3International Species Information System - ISIS, 2600 Eagan Woods Dr. Suite 50, MN 55121-1170, USA;Email: [email protected].

http://www.demographic-research.org 413

Kohler et al.: Comparative mortality levels among selected species of captive animals

1. Introduction

Authoritative accounts of survivorship and length of life exist for very few species. In thewild, uncertainty about the survival status of animals lost to follow up and imprecisionof age assignment are important hurdles to the accurate mapping of survival. Both inthe wild and in captivity, the small number of animals typically under observation entailssubstantial variability in estimates. Because so few data exist for any single species,comparisons of mortality across species are virtually non-existent.

The present study is an effort to begin addressing the deficiency of systematic data onspecies survival. This collection of life tables can also enrich data sets aimed at compilinglife history characteristics for various species (e.g. Ernest, 2003). The study utilizes alarge international data set, the International Species Information System (ISIS). ISIS wasfounded in 1973 as a network of approximately 50 North American and European zoos.The ISIS data base currently contains reports from approximately 650 zoos and aquariafrom over 70 countries on six continents. The database contains records of over 2 millionindividual zoo specimens. The selection of species for the present investigation was basedupon the specimens collection in the Smithsonian’s National Zoo in Washington D.C.,whose comparative mortality levels for recent years were examined in an unpublishedanalysis.

A species’ mortality profile in captivity can be expected to differ systematically fromthat in the wild. Courteney and Santow (1989) note that animals in captivity have thebenefits of veterinary care, a lack of predators, and a regular supply of food. On the otherhand, captive animals may suffer from higher levels of obesity (Taylor and Poole, 1998;Ward et al., 2003), injuries from exhibits (Leong et al., 2004), poor adaptation to captivityor to a zoo’s climate (Karstad and Sileo, 1971; Gozalo and Montoya, 1991) and frominbreeding that results in higher perinatal mortality (Wielebnowski, 1996). In addition,close quarters may facilitate the spread of infections, including those derived from otherspecies (Ward et al., 2003; de Wit, 1995). These relative risks will vary from species tospecies and from age to age, although the process of senescence may be similar in wildand captive populations of the same species (Hill et al., 2001).

2. Description of data and analytic scheme

We have selected 51 species for investigation. An individual animal is included in thedata set if it were a member of one of these species and if it were living or born in oneof the reporting zoos at any time during a recent period January 1, 1998—December31, 2003. Altogether, 35,229 animals are included in the data set. The record for eachanimal includes fields for date of birth, date of arrival at the zoo, date of death, and date

414 http://www.demographic-research.org

Demographic Research: Volume 15, Article 14

of departure from the zoo. It also includes information about sex, whether the animal wasborn in the wild or in captivity, and an assessment of the quality of information aboutbirth date.

The life tables that will be constructed are ‘period’ tables based upon age-specificmortality rates observed during the period from January 1, 1998 to December 31, 2003.Rather than following an actual cohort of births throughout life, a period life table followsa synthetic cohort of births and assumes that they are subject at each age to the age-specificdeath rates observed during a particular period (Preston et al., 2001).

Although the number of observations available on each species is larger than in nearlyall other studies, it is small enough that the estimation of age-specific death rates is subjectto substantial random error. Accordingly, we begin by creating large groups of speciesbased on order. We then examine how the mortality of each species relates to the aver-age mortality of the group. In our analyses, we treat species as independent statisticalunits. Species, however, are part of hierarchically structured phylogeny, and thus cannotbe regarded for statistical purposes as if drawn independently from the same distribu-tion (Felsenstein, 1985). Some methods have been proposed to circumvent this problemif adequate information on the phylogeny is available (Felsenstein, 1985). These analy-ses, however, are beyond the scope of the present paper. Any potential nonindependenceresulting from species sharing the same phylogeny in our analyses results in an under-estimation of the standard errors, and an overestimation of the statistical significance ofdifferences between species. The key findings of this paper that pertain to the generalmortality patterns of species, however, are unlikely to be affected by this nonindepen-dence.

Table 1 identifies the species under study, the number of individual animals contribut-ing observations to the analysis of survivorship, and the groupings that we have con-structed.

There are three types of entrance to the observational frame: birth in a reporting zooduring the period January 1, 1998—December 31, 2003; migration into a reporting zooduring this period; and survival in a reporting zoo past the beginning of the observationalperiod. Likewise, there are three sources of exit from the observational frame: deathduring the period; out-migration during the period; and survival past the terminal dateof the observational period (censoring). The events are reported according to the day,month, and year that they occurred, so that the exact number of animal-years contributedby each individual animal can be calculated. Age-specific death rates are constructed in aconventional fashion by counting the number of deaths in a particular age-time bloc anddividing that number by the exact number of animal-years lived in that age-time bloc.

Although detailed dates are reported for each event, they are not always accurate.The greatest risk of error pertains to birth date for animals born in the wild or whoseorigins were not recorded. Birth dates and ages typically must be estimated for these



Table 1: Species and groups of species investigated.

Number ofGroup name Species included animals

Apes Gorilla (Gorilla gorilla) 868Orangutan (Pongo pygmaeus) 675Siamang (Hylobates syndactylus) 357White-cheeked gibbon (Hylobates leucogenys) 169

Total 2,069

Small primates Ring-tailed lemur (Lemur catta) 2,545Ruffed lemur (Varecia variegata) 1,873Pygmy marmoset (Callithrix pygmaea) 1,310Colobus monkey (Colobus guereza) 960Geoffroy’s marmoset (Callithrix geoffroyi) 840Golden lion tamarin (Leontopithecus rosalia) 774Goeldi’s monkey (Callimico goeldi) 638Brown lemur (Eulemur fulvus) 576Golden-headed lion tamarin (Leontopithecus chrysomelas) 533Lion-tailed macaque (Macaca silenus) 507Sulawesi crested macaque (Macaca nigra) 348Howler monkey (Alouatta caraya) 269Dusky titi monkey (Callicebus moloch) 10

Total 11,183

Carnivores Lion (Panthera leo) 1,939Tiger (Panthera tigris) 1,626Cheetah (Acinonyx jubatus) 1,065Fennec fox (Vulpes zerda) 430Serval (Leptailurus serval) 422Bobcat (Lynx rufus) 406Fishing cat (Prionailurus viverrinus) 283Leopard cat (Prionailurus bengalensis) 257Caracal (Caracal caracal) 230Mexican grey wolf (Canis lupus baileyi) 227Spectacled bear (Tremarctos ornatus) 176Sloth bear (Melursus ursinus) 105New Guinea singing dog (Canis lupus hallstromi) 70Giant panda (Ailuropoda melanoleuca) 16

Total 7,252

(Continued)



Table 1 (Continued): Species and groups of species investigated.

Number ofGroup name Species included animals

Hoofstock North American bison (Bison bison) 1,810Arabian oryx (Oryx leucoryx) 1,079Reeves’s muntjac (Muntiacus reevesi) 954Przewalski’s wild horse (Equus caballus przewalskii) 666Grevy’s zebra (Equus grevyi) 552Dorcas gazelle (Gazella dorcas) 491Eld’s deer (Cervus eldi) 445Speke’s gazelle (Gazella spekei) 140

Total 6,137

Kangaroos Red kangaroo (Macropus rufus) 1,988Western grey kangaroo (Macropus fuliginosus) 581

Total 2,569

Crocodilians American alligator (Alligator mississippiensis) 1,914Johnston’s crocodile (Crocodylus johnstoni) 188Cuban crocodile (Crocodylus rhombifer) 67Indian gavial (Gavialis gangeticus) 33

Total 2,202

Ratites Greater rhea (Rhea americana) 1,383Common emu (Dromaius novaehollandiae) 1,207Cassowary (Casuarius casuarius) 213Darwin’s rhea (Pterocnemia pennata) 150

Total 2,953

Raptors Bald eagle (Haliaeetus leucocephalus) 649King vulture (Sarcorhamphus papa) 215

Total 864

Total Number of Animals 35,229



animals based upon age-characteristic behavior and morphology. When an animal withan uncertain birthdate enters a zoo, its birthdate will often be recorded as January 1.Table 2 shows the proportion of animals in each group with a stated birthdate of January 1.Overall, about one in every 24 animals is credited with a January 1 birth date. Clearly,the dating is least precise for crocodilians, where one in five is assigned a January 1 birthdate. Apes and small primates have the lowest percentage in this category, suggesting thatage assignment is unusually precise for them.

Table 2: Number of cases (N) and percentage of animals with birthdays reportedas January, 1st.

Group Name N PercentageApes 49 2.37Small primates 193 1.73Carnivores 196 2.70Hoofstock 200 3.26Kangaroos 114 4.44Crocodilians 412 18.71Ratites 208 7.04Raptors 75 8.68Total 1,447 4.11

A second area of data ambiguity also relates to the beginning of life. Analysis ofhuman longevity almost invariably begins with a live birth, typically defined followingWorld Health Organization criteria as an infant who shows any sign of life ([WHO] WorldHealth Organization, 1977). Fetal mortality is studied as a separate process. The require-ments of animal husbandry in zoos do not always admit a fine distinction between livebirths and stillbirths, and there are group-specific conventions regarding how to recordthe reproductive effort of different species. Among mammals, it is customary to includestillbirths among the births, with no distinction between live-born and stillborn offspring.Among birds and crocodilians, on the other hand, a birth is typically not counted un-til an egg is hatched. There are many instances in the ISIS database in which deaths arerecorded at age 0.00, a group that probably includes a mixture of stillbirths and live births.Life expectancy estimates are very sensitive to the estimation of the probabilities of deathamong the very young. In order to account for this sensitivity and to effect a partial dis-tinction between stillbirths and live births, we include in all life tables information onsurvivorship from age zero to age one week as well as survival beyond one week.



3. Results

3.1 Life tables for groups of species

To draw a preliminary picture of the characteristic mortality profiles of various largegroups of animals, we combine species in the fashion shown in Table 1. The calcula-tion of life tables for these groups is technically straightforward once a decision is madeabout how to ‘complete’ the life table by adopting an estimate of life expectancy at theoldest ages. A standard procedure is to invoke a relation characteristic of a stationary pop-ulation according to which life expectancy at the beginning of the open-ended age intervalis the reciprocal of the death rate above that age (Preston et al., 2001). One problem withthis approach in the present circumstance is that relatively few animals contribute obser-vations at very high ages and as result, the estimate of life expectancy at the oldest ages issubject to considerable sampling variability. In addition, the mortality rates at the oldestages may be biased by measurement problems such as age misreporting or the improperpresence in the data set of animals whose death at an earlier age was not registered.

An alternative approach to estimating mortality in the highest age interval is to fit astatistical model to data at younger ages and extrapolate the value of that function intothe oldest age interval. The most common statistical function used for this purpose is theGompertz curve (µ(x) = aebx), which specifies an exponential increase in the mortalityhazard with age. In our case the data are too sparse to allow us to distinguish confidentlyamong various competing mortality models. The Gompertz model is chosen because ofits simplicity (having only a level and a slope parameter), its familiarity, and the fact that itdoes a good job of representing adult mortality levels in a wide variety of species (Carneset al., 1996).

We have applied both approaches to estimate life expectancy at the age that begins theopen-ended interval. We select that age to be the lowest integer age above which fewerthan 2% of recorded deaths occur. The Gompertz function was fit to the individual-levelanimal data by using a standard maximum likelihood estimation procedure in STATA. Thecoefficient of correlation between the two series (N=7) is .89. This value is reassuringlyhigh in view of the fact that the methods of calculation are entirely independent of oneanother. Except for raptors, for whom the Gompertz model estimates a negative slope ofthe age-specific death rates, we use the Gompertz values hereafter because the model isparsimonious and provides accurate fit to the data.

Table 3 shows selected values from the life tables of these groups. Apes have thehighest life expectancy (expected number of additional years of life) at all ages and kan-garoos the lowest. Mortality levels in the first week of life are highly variable acrossgroups. Small primates and carnivores lose nearly 30% of births during this period, whilecrocodilians lose only 2.5%. As noted above, some of the differences in first-week mor-



tality among the groups are probably attributable to different ways of recording stillbirths.Nevertheless, the effect of variation in mortality during the first week of life on the rankordering of life expectancy is small. The ranking of groups from highest to lowest lifeexpectancy at one week of age is nearly the same as the ranking at age zero. By ageone, however, ratites have vaulted upwards in the life expectancy rankings as they escapeinfancy and their exceptionally high infant mortality rate.

Table 3: Selected life table parameters for groups of animals.

Number of animals (per 1,000 births) surviving to different ages1 week 1 year 2 years 5 years 10 years 15 years 20 years 25 years

Apes 837 747 727 700 657 602 552 473Small Primates 715 614 574 475 340 234 152Carnivores 713 568 534 480 384 211 64Hoofstock 862 732 672 577 431 295 180Kangaroos 936 762 666 485 267Crocodilians 975 836 751 634 542 470 399 314Ratites 870 525 463 382 270 205 144 94Raptors 778 689 663 590 471 393 353 315

Life expectancy (in years) at selected agesage 0 1 week 1 year 5 years 10 years 15 years

Apes 24.20 28.88 31.33 29.30 26.06 23.23Small Primates 8.45 11.80 12.70 11.84 10.58 9.29Carnivores 7.39 10.34 11.87 9.72 6.50 4.73Hoofstock 10.05 11.63 12.63 11.59 9.68 7.92Kangaroos 6.23 6.64 7.06 6.09 4.07 3.08Crocodilians 19.00 19.47 21.63 24.09 22.77 20.91Ratites 7.76 8.90 13.45 13.88 12.21 11.59Raptors 16.57 21.27 22.97 22.77 22.64 21.76

The age-pattern of mortality clearly differs among the groups. Beyond age 5, raptorsshow little sign of an increase in mortality. Life expectancy at 15 years is nearly the samefor them as it is at ages 1, 5, or 10. Crocodilians also show a slow diminution of lifeexpectancy with age. Other groups show a more conventional fall-off in life expectancyas age advances.

Groups that have high mortality at one age tend to have high mortality at other ages.Figure 1 plots the probability of surviving from age 1 to age 10 against the probabilityof surviving from age 10 to age 20 for the various groups. The coefficient of correlationbetween the two series is .71. Kangaroos, for whom the probability of survival to age 20is derived from the Gompertz model, are lowest on both indicators and apes are highest.



Other groups generally fall in an orderly intermediate position with the exception of car-nivores, who have moderate mortality between ages one and ten and quite high mortalitybetween ages 10 and 20.

Figure 1: Scatterplot of survival probabilities within two age intervals for groupsof animals

primates

apes

carnivores

hoofstock

crocodilians

ratites

raptors

kangaroos

0.2

.4.6

.8P

roba

bilit

y of

sur

vivi

ng fr

om a

ge 1

0 to

age

20

.4 .6 .8 1Probability of surviving from age 1 to age 10

Survival Probabilities within Two Age Intervals for Groups of Animals

As theory predicts and as will be demonstrated below, mortality in these heteroge-neous groups rises less rapidly with age than it typically does in the individual speciesthat make up the groups. The most longevous species become relatively more prevalentwith age, imparting a downward bias to the age-slope of mortality for the group as a whole(Vaupel and Yashin, 1985).

3.2 Mortality variation by species, sex, and birth type

In order to identify how mortality varies by species within groups, we assume that theeffect of a particular covariate such as sex or taxonomy has the same proportional effecton the hazard of death at all ages beyond five within the group. We estimate a non-parametric piecewise-constant proportional hazard model, with the mortality hazard at



age x specified asµ(x) = h0(x) exp(βzi), (1)

where h0(x) is the piecewise-constant specification of the baseline hazard with constantmortality risks within one year age intervals, and zi are binary variables for sex, birth typeor taxonomy of the animal.

The proportional hazard approach enables us to identify how the mortality of eachspecies relates to that of all other species within the group. Other covariates availablein the data set that can also be examined are sex and animal’s place of birth (i.e., wild,captivity or unknown place of birth). We estimate the hazard models from age 5 becausethe proportionality assumption seems less likely to hold across younger ages.

The reference categories that we choose for this analysis (the categories to which allother categories will be compared) are males, animals born in captivity, and the speciesthat contributes the largest number of observations within a group. Animals of unknownsex are excluded from the analysis by sex, but included in other model specifications.

Table 4 shows the coefficients pertaining to sex. Four of the eight groups—apes, hoof-stock, crocodilians, and kangaroos—demonstrate significant sex differences in mortalitybeyond age five. In each of these cases, females have mortality that is lower than that ofmales by 33-40%. Sex differences are smaller and insignificant in the remaining groups.Previous studies of sex differences in mortality have typically been based upon fewer ob-servations and indirect indicators, such as the sex ratio of the living population (Taylorand Poole, 1998; Hill et al., 2001; Promislow, 1992). Our results offer partial supportto a generalization that has emerged from this literature that males typically have highermortality among mammals (Promislow, 1992).

There are several reports in the literature of animals who suffer excess mortality whenbrought into a zoo, relative to animals born and reared in the zoo (Wallace et al., 1987;Tocidlowski et al., 1997; Gozalo and Montoya, 1991). Emotional trauma and exposure tonew diets and disease environments are among the causes cited. Table 4 also presents themortality levels of wild-born animals relative to those of captive-born animals. In noneof the groups is the effect of being wild-born statistically significant, and in four of theeight groups the mortality of wild-born animals is lower than that of zoo-born animals.Excess mortality of wild-born animals brought to zoos does not appear to be a generalphenomenon. Nevertheless, it should be noted that many of the wild-born animals werebrought to a zoo well before the period of observation began. For these animals, anyimmediate momentary trauma associated with the move would not be reflected in theresults.

Figures 2 and 3 show the coefficients expressing the level of mortality at ages 5 andbeyond for each species, relative to the species chosen as the reference category within acertain group. The hazard ratios are plotted on a logarithmic scale to reflect the relative



Table 4: Hazard ratios for sex and birth type obtained from piecewise-constanthazard models for groups of animals.

Hazard ratio for Hazard ratio forFemales Birth Type

Reference category males born in captivityUnknown Wild born

Apes 0.662** 1.522 1.469(0.097) (0.538) (0.350)

Small primates 1.088 1.026 1.110(0.057) (0.182) (0.170)

Carnivores 0.945 0.912 0.887(0.054) (0.131) (0.096)

Hoofstock 0.620** 1.058 1.294(0.048) (0.281) (0.418)

Kangaroos 0.670** 0.340** 0.948(0.078) (0.141) (0.219)

Crocodilians 0.609* 0.961 1.035(0.142) (0.276) (0.318)

Ratites 1.255+ 1.186 0.811(0.154) (0.218) (0.328)

Raptors 0.953 0.768 0.638(0.233) (0.360) (0.222)

Notes: Standard errors in parentheses. p-values: + p < 0.05; * p < 0.01; ** p < 0.001.

size of bars corresponding to coefficients larger and smaller than 1. Figure 2 shows thatmortality variation is enormous within the group of smaller primates. Mortality ratesvary by a factor of nearly eight between brown lemurs at the low extreme and pygmymarmosets at the high. Tamarins are intermediate between the lemurs and the marmosets.

In contrast, apes show relatively little differentiation in their post-5 mortality levels,although siamangs have significantly higher mortality than gorillas. Orangutans havemortality that exceeds that of gorillas by 30% but the difference is not statistically sig-nificant. The range of variation is also small among hoofstock. North American bisons,the reference category, have the lowest mortality and gazelles the highest. The group ofcarnivores shown in Figure 3 also presents a highly varied picture, with bears having verylow mortality and cheetahs very high compared to the reference category tigers. Lionsand tigers are virtually identical to one another in their post-5 mortality levels.

Significant differences in mortality are found also among kangaroos (not shown here)with grey kangaroos having about 33% lower risk of death than red kangaroos. Differ-



Figure 2: Hazard ratios obtained from piecewise-constant model for primates,apes and hoofstock.

haza

rd r

atio

0.52**0.58**

0.81* 0.85

Ref

1.21+

1.68**1.8**

2.31**

2.74**

3.25*

3.97**4.45**

Lion

−ta

iled

mac

aque

Bro

wn

lem

ur

Ruf

fed

lem

ur

How

ler

mon

key

Rin

g−ta

iled

lem

ur

Col

obus

mon

key

Gol

den

lion

tam

arin

Sul

awes

i cr.

mac

aque

G.−

head

ed ta

mar

in

Goe

ldi’s

mon

key

Dus

ky ti

ti m

onke

y

Geo

ffroy

’s m

arm

oset

Pyg

my

mar

mos

et

0.5

0.67

11.

52

34

5

(a) Primates

haza

rd r

atio

Ref

1.15

1.3

1.67*

Gor

illa

Whi

te−

ch. g

ibbo

n

Ora

ngut

an

Sia

man

g

11.

5

(b) Apes

haza

rd r

atio

Ref

1.25+1.31+

1.96**

2.81**2.96**

3.48**

6.93**

Nor

th A

m. b

ison

Prz

ewal

ski’s

hor

se

Gre

vy’s

zeb

ra

Ara

bian

ory

x

Ree

ves’

s m

untja

c

Eld

’s d

eer

Dor

cas

gaze

lle

Spe

ke’s

gaz

elle

11.

52

34

56

78

(c) Hoofstock

Note: p-values: + p < 0.10; * p < 0.05; ** p < 0.01.



Figure 3: Hazard ratios obtained from piecewise-constant model for carnivores,crocodilians and ratites.

haza

rd r

atio

0.18**

0.26**

0.34

0.5**

0.62**

0.97

Ref 1.05

1.42+1.66* 1.73**

3.1**3.39** 3.64**

Spe

ctac

led

bear

New

Gui

nea

dog

Gia

nt p

anda

Slo

th b

ear

Bob

cat

Lion

Tig

er

Ser

val

Mex

ican

wol

f

Leop

ard

cat

Car

acal

Che

etah

Fis

hing

cat

Fen

nec

fox

0.14

0.25

0.5

11.

52

34 (d) Carnivores

haza

rd r

atio

0.880.89

Ref

1.5

Cub

an c

roco

dile

Indi

an g

avia

l

Am

eric

an a

lliga

tor

John

ston

’s c

roco

dile

11.

5 (e) Crocodilians

haza

rd r

atio

Ref

1.13

1.9**

4.79**

Com

mon

em

u

Cas

sow

ary

Gre

ater

rhe

a

Dar

win

’s r

hea

11.

52

34

5

(f) Ratites

Note: p-values: + p < 0.10; * p < 0.05; ** p < 0.01.



ences in mortality persist also by species among ratites and raptors. Among the non-flyingbirds (ratites), rheas have the highest mortality levels, and among raptors (not shown here)king vultures have much lower mortality than bald eagles. In summary, significant differ-ences in mortality persist between species within nearly all groups except crocodilians.

3.3 Life table parameters for individual species

In this section we present life table functions for 42 species. This number is lower than thenumber of species considered in previous sections for the following reasons: a) Crocodil-ians do not show significant inter-species mortality variation, so the four species havebeen combined into one life table; b), there were too few observations among pandas,New Guinea singing dogs, and dusky titi monkeys (N below 75 in each case) to permitreliable tables to be constructed; c), 98% of the specimens of Darwin’s rhea died beforereaching age one, and d), king vultures and howler monkeys had too few observations atolder ages to allow life tables to be properly terminated.

Below age five, the life tables that we construct are based upon directly observeddeath rates for a species in single years of age (with the first week of life distinguished ininfancy). Beyond age five, we used one of two strategies for constructing a life table: con-ventional application of directly observed death rates in single years of age for the species(i.e., the same approach used below age five); or use of the hazard model results for thespecies. The hazard model approach was employed for the 15 species identified in Table 5for which the direct use of death rates produced erratic age-patterns of mortality. When di-rect data were used, life tables were terminated in the conventional fashion by estimatinglife expectancy as the reciprocal of the age-specific death rate in the open-ended interval(Preston et al., 2001). When the hazard model was used, a species’ estimated hazard wasassumed to apply in the open-ended interval as well as in all other ages beyond five. Inthis case, the shape of the hazard was estimated for the group of which the species was amember, and the level of the hazard for the species in question was estimated through thehazard model approach described above. Although results beyond age five are shown infive-year intervals, all calculations are performed in single-year age intervals.

Table 5 presents two life table functions for each species: a) the probability of surviv-ing from birth to various subsequent ages (or between any pair of ages displayed); b) andthe expected number of years of additional life for animals who have reached a particularage. Among the species shown in Table 5, the North American bison has the longest lifeexpectancy at birth (15.1 years) followed by the lion-tailed macaque (13.4 years) and thebald eagle and the common emu (both 13.2 years). The fennec fox has the shortest lifeexepctancy at birth among the species shown in Table 5 (2.6 years). By one week of age,the spectacled bear (life expectancy of 23.9 years) has replaced the North American bisonas the most longevous species. The spectacled bear is, however, characterized by a very



Table 5: Life expectancy and survival functions for various species.

Age0 1 week 1 2 3 4 5 10 15 20 25 30 35

Small PrimatesGoeldi’s monkey* ex 6.0 8.5 8.4 8.3 7.9 7.4 6.9 5.3 4.1(Callimico goeldi) lx 1.00 0.70 0.63 0.57 0.53 0.49 0.46 0.26 0.12Geoffroy’s marmoset ex 4.3 6.1 6.7 6.4 6.1 5.8 5.6 4.7(Callithrix geoffroyi) lx 1.00 0.71 0.55 0.50 0.44 0.39 0.34 0.15Pygmy marmoset* ex 3.6 5.6 6.0 5.9 5.5 5.2 5.0 4.1(Callithrix pygmaea) lx 1.00 0.65 0.50 0.44 0.39 0.34 0.29 0.12Colobus monkey ex 10.1 12.4 12.8 12.5 12.0 11.6 11.5 9.6 7.4 5.0 2.5(Colobus guereza) lx 1.00 0.82 0.73 0.69 0.66 0.63 0.59 0.44 0.31 0.20 0.09Brown lemur* ex 12.9 18.2 20.4 19.6 18.6 17.6 17.2 14.0 11.1 8.5 5.9(Eulemur fulvus) lx 1.00 0.71 0.60 0.60 0.60 0.60 0.58 0.51 0.43 0.34 0.24Ring-tailed lemur ex 12.1 14.3 15.3 15.0 14.8 14.3 13.7 11.9 9.9 8.6(Lemur catta) lx 1.00 0.84 0.73 0.71 0.67 0.65 0.62 0.49 0.37 0.25Golden-h. lion tamarin ex 5.4 8.3 8.5 8.2 8.3 7.9 7.8 5.4 4.2(Leontopithecus

chrysomelas) lx 1.00 0.65 0.57 0.52 0.46 0.42 0.38 0.25 0.11Golden lion tamarin ex 5.1 10.6 11.0 10.9 10.9 10.6 10.2 6.6 3.9 2.1(Leontopithecus rosalia) lx 1.00 0.47 0.42 0.39 0.35 0.33 0.31 0.26 0.17 0.06Sulawesi cr. macaque* ex 7.6 10.7 11.1 10.3 10.0 9.7 9.2 7.2 5.5 4.0 2.6(Macaca nigra) lx 1.00 0.71 0.62 0.61 0.57 0.53 0.51 0.35 0.21 0.10 0.03Lion-tailed macaque ex 13.4 19.2 20.0 19.2 19.4 18.9 18.4 14.7 12.4 9.3 5.1 6.7 4.7(Macaca silenus) lx 1.00 0.69 0.64 0.63 0.59 0.58 0.56 0.52 0.43 0.36 0.33 0.09 0.04Ruffed lemur ex 10.8 16.6 17.5 17.2 16.6 16.2 15.3 12.6 10.6 8.8(Varecia variegata) lx 1.00 0.65 0.58 0.56 0.55 0.53 0.52 0.44 0.34 0.25

CarnivoresCheetah ex 6.4 8.1 9.1 8.5 7.6 7.0 6.4 3.3 1.4(Acinonyx jubatus) lx 1.00 0.79 0.63 0.60 0.59 0.56 0.53 0.35 0.08Mexican grey wolf ex 8.4 11.8 12.5 11.5 10.5 9.5 8.9 5.4 8.8(Canis lupus baileyi) lx 1.00 0.71 0.62 0.62 0.62 0.62 0.59 0.48 0.11Caracal* ex 7.0 9.6 11.4 10.4 9.5 9.1 8.3 4.9 2.8(Caracal caracal) lx 1.00 0.72 0.55 0.55 0.55 0.52 0.50 0.39 0.17Serval* ex 9.5 12.4 13.2 12.2 11.7 11.0 10.2 6.4 3.7 2.4(Leptailurus serval) lx 1.00 0.77 0.67 0.67 0.64 0.63 0.61 0.53 0.32 0.09Bobcat ex 11.5 14.0 15.9 15.1 14.5 13.5 12.8 8.6 5.1 2.9(Lynx rufus) lx 1.00 0.82 0.67 0.67 0.65 0.65 0.63 0.59 0.46 0.20Sloth bear* ex 8.2 16.3 18.1 17.1 16.1 15.1 14.1 10.0 6.9 5.6 4.7 3.8 0.8(Melursus ursinus) lx 1.00 0.50 0.43 0.43 0.43 0.43 0.43 0.40 0.31 0.17 0.08 0.04 0.02Lion ex 6.4 9.0 11.2 11.9 11.9 11.0 10.5 6.9 3.9 1.7(Panthera leo) lx 1.00 0.71 0.52 0.45 0.41 0.41 0.39 0.33 0.22 0.07Tiger ex 9.6 13.3 13.4 12.8 12.0 11.4 10.6 6.5 3.6 2.7(Panthera tigris) lx 1.00 0.72 0.66 0.64 0.63 0.61 0.60 0.54 0.35 0.08Leopard cat ex 5.3 7.0 9.4 9.3 8.7 8.7 8.7 5.5 3.5(Prionailurus

bengalensis) lx 1.00 0.75 0.49 0.45 0.43 0.38 0.34 0.26 0.13Fishing cat ex 6.1 8.5 9.4 8.4 7.9 7.2 6.2 2.7 1.1(Prionailurus viverrinus) lx 1.00 0.71 0.58 0.58 0.54 0.52 0.52 0.37 0.04(Continued)



Table 5 (Continued): Life expectancy and survival functions for various species.

Age0 1 week 1 2 3 4 5 10 15 20 25 30 35

Spectacled bear ex 12.8 23.9 24.8 23.8 22.8 21.8 20.8 15.8 11.8 7.2 3.5 4.1(Tremarctos ornatus) lx 1.00 0.53 0.49 0.49 0.49 0.49 0.49 0.49 0.46 0.44 0.34 0.06Fennec fox ex 2.6 5.4 8.3 7.8 7.2 6.4 5.7 2.6(Vulpes zerda) lx 1.00 0.48 0.27 0.25 0.24 0.23 0.22 0.13

HoofstockNorth American bison* ex 15.1 16.6 17.4 18.0 17.5 17.6 16.9 14.4 12.4 11.7(Bison bison) lx 1.00 0.91 0.82 0.75 0.73 0.68 0.67 0.57 0.46 0.32Eld’s deer* ex 5.1 8.3 9.4 9.1 8.6 8.4 7.7 5.9 4.0(Cervus eldi) lx 1.00 0.61 0.48 0.45 0.42 0.38 0.37 0.23 0.12Przewalski’s wild horse* ex 12.9 15.0 14.5 15.2 14.6 13.9 13.2 10.5 8.1 6.3 4.3 3.2(Equus caballus

przewalskii) lx 1.00 0.86 0.83 0.74 0.72 0.71 0.69 0.57 0.43 0.27 0.16 0.05Grevy’s zebra* ex 11.0 13.4 14.2 14.3 14.1 13.5 12.8 10.2 7.8 6.1 4.2 3.1(Equus grevyi) lx 1.00 0.82 0.72 0.67 0.63 0.61 0.60 0.48 0.36 0.22 0.13 0.04Dorcas gazelle* ex 5.3 6.3 8.0 7.7 7.6 7.1 7.3 5.8(Gazella dorcas) lx 1.00 0.84 0.58 0.53 0.47 0.44 0.37 0.21Speke’s gazelle* ex 5.2 5.5 5.4 5.5 5.0 4.4 4.4 3.7(Gazella spekei) lx 1.00 0.94 0.80 0.66 0.59 0.55 0.44 0.15Reeves’s muntjac* ex 8.2 9.6 10.4 10.2 10.0 9.3 8.6 7.1 6.1(Muntiacus reevesi) lx 1.00 0.85 0.71 0.66 0.61 0.59 0.57 0.37 0.19Arabian oryx ex 10.1 10.8 11.7 11.4 11.0 10.5 9.9 6.9 4.4(Oryx leucoryx) lx 1.00 0.93 0.78 0.74 0.70 0.67 0.64 0.50 0.32

RatitesCassowary ex 11.6 17.4 19.4 18.9 18.3 17.6 16.6 14.7 12.1 10.1 6.8 3.5(Casuarius casuarius) lx 1.00 0.66 0.56 0.55 0.54 0.53 0.53 0.43 0.36 0.28 0.23 0.16Common emu ex 13.2 14.1 18.1 18.7 18.4 18.1 17.5 15.5 13.8 11.3 11.0 9.1(Dromaius

novaehollandiae) lx 1.00 0.94 0.69 0.63 0.61 0.58 0.57 0.48 0.38 0.31 0.21 0.15Greater rhea ex 5.5 6.4 10.8 11.4 11.4 11.2 11.1 9.2 8.6(Rhea americana) lx 1.00 0.85 0.45 0.39 0.35 0.33 0.30 0.22 0.14

RaptorsBald eagle ex 13.2 17.3 19.0 18.9 19.0 19.0 18.8 18.8 17.3 14.2 10.8 8.7 7.6(Haliaeetus

leucocephalus) lx 1.00 0.76 0.66 0.63 0.59 0.56 0.54 0.41 0.34 0.31 0.27 0.20 0.12

KangaroosWestern grey kangaroo* ex 7.3 7.8 8.5 8.7 8.2 7.9 7.7 5.5 4.8(Macropus fuliginosus) lx 1.00 0.93 0.76 0.66 0.62 0.57 0.52 0.33 0.15Red kangaroo ex 6.0 6.4 6.8 6.7 6.5 6.2 5.9 4.0(Macropus rufus) lx 1.00 0.94 0.76 0.67 0.59 0.53 0.47 0.25

Notes: * Parameters after age 5 are based upon piecewise-constant hazard model.



high mortality at or shortly after birth and only 53% of the animals survive to one weekof age. A similar pattern is observed for the sloth bear and the fennec fox among whomonly half of the born cubs survive the first week of their life. Among small primates, thegolden-lion tamarin is characterized by the highest mortality between birth and one weekof age. Lions and tigers have very similar mortality beyond age three although lions havehigher death rates below that age.

Not all remaining species are shown in Table 5 because several important species donot exhibit enough deaths at older ages during the period under study to enable the lifetable to be properly completed. This pattern is observed among crocodilians and apes.In these species, the recorded death rates during the 1998-2003 period imply that largenumbers would survive to older ages. However, in the zoo collections during the periodunder investigation, there are too few deaths at those older ages to allow the life table tobe confidently closed out. For example, the life table for gorillas indicates that 31% willsurvive to age 40, but in the populations and period under study there were only 146 yearsof exposure and 9 deaths recorded above age 40.

There are several possible explanations of this anomaly: a) Death rates for a specieswere unusually low during the period 1998-2003; the shortage of very old specimensduring this period is a result of higher death rates during earlier periods. b) Deaths areunderrecorded; c) The number of specimens is overestimated. This latter problem wouldbe particularly serious if specimens who had died or out-migrated were erroneously main-tained on a zoo’s books. In this case, the fraction of estimated years at risk that consistedof absent animals would grow as age advances, and death rates at older ages would beprogressively biased downwards.

We consider it unlikely that deaths in these species were underrecorded because fourof the five species affected by the problem are apes who are often among the most visibleand important species in a collection. In the case of gorillas, it seems more likely thatthe ISIS data base includes inaccurate records of survivors. Gorillas are assigned multiplestudbook numbers in various regions around the world, which is a source of confusionin the record system (Flesness et al., 1995). One analysis argues that "...the mish moshof studbook IDs that presently permeates the ISIS database for gorillas does not permiteither genetic or demographic analyses for one of the highest profile and most charismaticmegavertabrates in captivity." (Earnhardt et al., 1995).

Rather than discarding all information for these important species, we have chosenan age at which to terminate the life table and present survival information up to thatage. Thus, rather than the conventional life expectancy values, we present ‘partial’ lifeexpectancy values that refer to the expected number of years to be lived before the selectedterminal age. That age is 40 in the case of gorillas and orangutans, 35 for siamangs, and30 for gibbons and crocodilians. The fact that there is an internal inconsistency between



the recorded survival for these species and the number of older specimens in collectionsmeans that the results, shown in Table 6, are less reliable than for other species presented.

Table 6: Partial life expectancy and survival functions for Apes and Crocodilians.

Age0 1 week 1 2 3 4 5 10 15 20 25 30 35 40

ApesGorilla ex 23.3 27.2 29.1 28.7 28.1 27.1 26.4 22.7 19.1 15.3 12.1 8.3 4.4 0*(Gorilla gorilla) lx 1.00 0.86 0.77 0.76 0.75 0.75 0.74 0.70 0.65 0.61 0.54 0.48 0.40 0.31*White-ch. gibbon ex 15.3 20.4 21.9 22.8 21.8 20.8 20.5 16.6 12.8 8.7 4.7 0*(Hylobates

leucogenys) lx 1.00 0.75 0.67 0.61 0.61 0.61 0.59 0.56 0.51 0.47 0.41 0.36Siamang ex 19.5 22.2 23.7 23.0 22.3 21.3 21.0 17.7 14.6 11.1 8.2 4.4 0*(Hylobates

syndactylus) lx 1.00 0.88 0.79 0.78 0.77 0.77 0.74 0.68 0.60 0.54 0.43 0.36 0.26*Orangutan ex 19.0 23.6 26.3 26.0 25.3 25.5 24.5 21.0 17.8 14.3 11.4 7.9 4.3 0*(Pongo pygmaeus) lx 1.00 0.80 0.69 0.67 0.67 0.64 0.64 0.59 0.54 0.50 0.42 0.36 0.28 0.20*Crocodilians ex 14.6 14.9 16.4 17.2 17.9 17.7 17.1 14.6 11.5 8.1 4.5 0*

lx 1.00 0.97 0.84 0.75 0.68 0.65 0.63 0.54 0.47 0.40 0.31 0.25*

Notes: * By assumption; The group of crocodilians includes American alligator, Johnston’s crocodile, Cuban

crocodile and Indian gavial.

Despite the assumption that no years would be lived beyond age 40 by a cohort ofnewborn gorillas, the gorilla life expectancy at birth, 23.3 years, is the highest of anyspecies considered. Likewise, orangutans, gibbons, and siamangs achieve a higher lifeexpectancy at birth than any other species under review despite having their survivorshipartificially truncated. These species of apes also provide four out of the five highest lifeexpectancies at one week of age (with spectacled bears). Clearly, apes appear to livelonger in captivity than the other species that we have reviewed, but data inconsistenciesadd uncertainty to this conclusion.

4. Discussion

It is useful to compare the estimates of life expectancy that we have produced to other esti-mates of longevity. The largest collection of longevity records for vertebrate species/sub-species is contained in Carey and Judge (2000). Rather than tabulating the average lengthof life for a species, Carey and Judge tabulated the highest age recorded to have been at-tained by a member of that species. Unlike the values presented in this paper, the highestobserved age will depend systematically upon the number of animals observed, whichintroduces noise into the Carey and Judge estimates.



Figure 4 compares our estimates of life expectancy at age one week for mammalianspecies to estimates of the highest age attained by that species (N=37). Life expectancyat age one week is used rather than at age zero in order to minimize the influence of fetalmortality. The relationship is tight, with a coefficient of correlation between the two seriesof .87. The line fitted on the graph indicates that the average length of life of a speciesis approximately equal to half of the highest age attained by that species. For the manymammalian species not included in this analysis and for whom no reliable life table hasbeen constructed but for whom a maximum attained age has been estimated, this rule ofthumb may prove useful.

Figure 4: Relationship between life expectancy and highest age ever attained bymammalian species.

However, non-mammalian species show a much poorer fit between these two vari-ables, with a correlation of only .55 (N=5). It is possible that the fit is poorer becauseCarey and Judge were unable to observe as many members of these species as of mam-mals, adding greater variability to their estimates. For mammals and non-mammals com-bined, the correlation coefficient is .77.

Having a large group of species for whom life tables have been prepared using com-



parable data and methods has enabled us to investigate a number of issues that cannotbe considered when one species is considered in isolation. These include questions ofthe generality of sex differences in mortality and of the excess mortality of wild-borncompared to zoo-born animals. The comparative design has also enabled us to generalizeabout the relation between the life expectancy of a species and the highest age attainedby that species. Although we have not pursued the connections here, the estimates canalso be used as the basis for studying the relation between longevity and age at sexualmaturity, length of the reproductive period, length of juvenile dependency, and patternsof physical growth. The data base of the International Species Information System is anextremely valuable resource for investigations of the life cycle within and across species.



References

Carey J R, Judge D S. 2000. Longevity Records: Life Spans of Mammals, Birds, Am-phibians, Reptiles, and Fish. Odense: Monographs on Population Aging 8, OdenseUniversity Press.

Carnes B A, Olshansky S J, Grahn D. 1996. Continuing the search for a law of mortality.Population and Development Review 22:231–264.

Courteney J, Santow G. 1989. Mortality of wild and captive chimpanzees. Folia Prima-tologica 52:167–177.

de Wit J J. 1995. Mortality of rheas caused by a synchamus trachea infection. VeterinaryQuarterly 17:39–40.

Earnhardt J M, Thompson S D, Willis K. 1995. Reply to Flessness et al. Zoo Biology14:519–522.

Ernest, S K. Morgan 2003. Life History Characteristics of Placental Nonvolant Mammals.Ecology 84:3402.

Flesness N R, Lukens D R, Porter S B, Wilson C R, Grahn L V. 1995. ISIS and stud-books, very high census correlation for the north american zoo population: a replyto Earnhardt, Thompson, and Willis. Zoo Biology 14:509–517.

Felsenstein, J. 1985. Phylogenies and the Comparative Method. The American Naturalist125:1-15.

Gozalo A, Montoya E. 1991. Mortality causes of the moustached tamarin (Saguinusmystax) in captivity. Journal of Medical Primatology 21:35–38.

Hill K, Boesch C, Goodall J A, Pusey A, Williams J, Wrangham R. 2001. Mortality ratesamong wild chimpanzees. Journal of Human Evolution 40(5): 437–450.

Karstad L, Sileo L. 1971. Causes of death in captive wild waterfowl in the KortrightWaterfowl Park, 1967-1970. Journal of Wildlife Diseases 7:236–241.

Leong K M, Terrell S P, Savage A. 2004. Causes of mortality in captive cotton-toptamarins (Saguinus oedipus). Zoo Biology 23:127–137.

Preston S H, Heuveline P., Guillot M. 2001. Demography: Measuring and ModelingPopulation Processes. Oxford: Blackwell Publishers.

Promislow D E L. 1992. Costs of sexual selection in natural populations of mammals.Proceedings: Biological Sciences 247:203–210.



Taylor V J, Poole T B. 1998. Captive breeding and infant mortality in Asian elephants; Acomparison between twenty Western zoos and three Eastern elephant centers. ZooBiology 17:311–332.

Tocidlowski M E, Cornish T E, Loomis M R, Stoskopf M K. 1997. Mortality in cap-tive wild-caught horned puffin chicks (Fratercula Corniculata). Journal of Zoo andWildlife Medicine 28:298–306.

Vaupel J W, Yashin A I. 1985. Heterogeneity’s ruses: Some surprising effects of selectionon population dynamics. American Statistician 39:176–185.

Wallace R S, Bush M, Montali R J. 1987. Deaths from exertional myopathy at the NationalZoological Park from 1975 to 1985. Journal of Wildlife Diseases 23:454–462.

Ward M P, Ramer J C, Proudfoot J, Garner M M, Juan-Sallès C, Wu C C. 2003. Outbreakof salmonellosis in a zoologic collection of Lorikeets and Lories (Trichoglossus,Lorius, and Eos spp.). Avian Diseases 47:493–498.

[WHO] World Health Organization. 1977. Manual of Mortality Analysis. Geneva: Divi-sion of Health Statistics, World Health Organization.

Wielebnowski N. 1996. Reassessing the relationship between juvenile mortality and ge-netic monomorphism in captive cheetahs. Zoo Biology 15:353–369.


Documents

Comparitive mortality levels among selected … · cantly exceeds that of males above age ﬁve. Wild-born animals do not have mortality that differs signiﬁcantly from captive-born